Evidence Synthesis for Decision Making in Healthcare

Description

In the evaluation of healthcare, rigorous methods of quantitative assessment are necessary to establish interventions that are beneficial, are superior to all alternatives and are cost-effective. Usually one study will not provide answers to these questions and it will be necessary to synthesize evidence from multiple sources. This book aims to outline a coherent approach to such evidence synthesis, for the purpose of decision making. Each chapter contains worked examples, exercises and solutions drawn from a variety of medical disciplines

Evidence Syntesis for Decision Making intends to provide a practical guide to the appropriate methods for synthesizing evidence for use in analytical decision models. More specifically, it proposes a comprehensive evidence synthesis framework, which models all the available data appropriately and efficiently in a format that can be incorporated directly into a decision model.

Table of Contents

• Preface

1. Introduction
1. The rise of health economics
2. Decision making under uncertainty
1. Deterministic models
2. Probabilistic decision modelling
3. Evidence-based medicine
4. Bayesian statistics
5. NICE
6. Structure of the book
7. Summary key points
8. Further reading
• References
2. Bayesian methods and WinBUGS
1. Introduction to Bayesian methods
1. What is a Bayesian approach?
2. Likelihood
3. Bayes’ theorem and Bayesian updating
4. Prior distributions
5. Summarising the posterior distribution
6. Prediction
7. More realistic and complex models
8. MCMC and Gibbs sampling
2. Introduction to WinBUGS
1. The BUGS language
2. Graphical representation
3. Running WinBUGS
4. Assessing convergence in WinBUGS
5. Statistical inference in WinBUGS
6. Practical aspects of using WinBUGS
3. Advantages and disadvantages of a Bayesian approach
4. Summary key points
5. Further reading
6. Exercises
• References
3. Introduction to decision models
1. Introduction
2. Decision tree models
3. Model parameters
1. Effects of interventions
2. Quantities relating to the clinical epidemiology of the clinical condition being treated
3. Utilities
4. Resource use and costs
4. Deterministic decision tree
5. Stochastic decision tree
1. Presenting the results of stochastic economic decision models 60
6. Sources of evidence
7. Principles of synthesis for decision models (motivation for the rest of the book)
8. Summary key points
9. Further reading
10. Exercises
• References
4. Meta-analysis using Bayesian methods
1. Introduction
2. Fixed Effect model
3. Random Effects model
1. The predictive distribution
2. Prior specification for τ
3. ‘Exact’ Random Effects model for Odds Ratios based on a Binomial likelihood
4. Shrunken study level estimates
4. Publication bias
5. Study validity
6. Summary key points
7. Further reading
8. Exercises
• References
5. Exploring between study heterogeneity
1. Introduction
2. Random effects meta-regression models
1. Generic random effect meta-regression model
2. Random effects meta-regression model for Odds Ratio (OR) outcomes using a Binomial likelihood
3. Autocorrelation and centring covariates
3. Limitations of meta-regression
4. Baseline risk
1. Model for including baseline risk in a meta-regression on the (log) OR scale
2. Final comments on including baseline risk as a covariate
5. Summary key points
6. Further reading
7. Exercises
• References
6. Model critique and evidence consistency in random effects meta-analysis
1. Introduction
2. The Random Effects model revisited
3. Assessing model fit
1. Deviance
2. Residual deviance
4. Model comparison
1. Effective number of parameters, pD
2. Deviance Information Criteria
5. Exploring inconsistency
1. Cross-validation
2. Mixed predictive checks
6. Summary key points
7. Further reading
8. Exercises
• References
7. Evidence synthesis in a decision modelling framework
1. Introduction
2. Evaluation of decision models: One-stage vs two-stage approach
3. Sensitivity analyses (of model inputs and model specifications)
4. Summary key points
5. Further reading
6. Exercises
• References
8. Multi-parameter evidence synthesis
1. Introduction
2. Prior and posterior simulation in a probabilistic model: Maple Syrup Urine Disease (MSUD)
3. A model for prenatal HIV testing
4. Model criticism in multi-parameter models
5. Evidence-based policy
6. Summary key points
7. Further reading
8. Exercises
• References
9. Mixed and indirect treatment comparisons
1. Why go beyond ‘direct’ head-to-head trials?
2. A fixed treatment effects model for MTC
1. Absolute treatment effects
2. Relative treatment efficacy and ranking
3. Random Effects MTC models
4. Model choice and consistency of MTC evidence
1. Techniques for presenting and understanding the results of MTC
5. Multi-arm trials
6. Assumptions made in mixed treatment comparisons
7. Embedding an MTC within a cost-effectiveness analysis
8. Extension to continuous, rate and other outcomes
9. Summary key points
10. Further reading
11. Exercises
• References
10. Markov models
1. Introduction
2. Continuous and discrete time Markov models
3. Decision analysis with Markov models
1. Evaluating Markov models
4. Estimating transition parameters from a single study
1. Likelihood
2. Priors and posteriors for multinomial probabilities
5. Propagating uncertainty in Markov parameters into a decision model
6. Estimating transition parameters from a synthesis of several studies
1. Challenges for meta-analysis of evidence on Markov transition parameters
2. The relationship between probabilities and rates
3. Modelling study effects
4. Synthesis of studies reporting aggregate data
5. Incorporating studies that provide event history data
6. Reporting results from a Random Effects model
7. Incorporating treatment effects
7. Summary key points
8. Further reading
9. Exercises
• References
11. Generalised evidence synthesis
1. Introduction
2. Deriving a prior distribution from observational evidence
3. Bias allowance model for the observational data
4. Hierarchical models for evidence from different study designs
5. Discussion
6. Summary key points
7. Further reading
8. Exercises
• References
12. Expected value of information for research prioritisation and study design
1. Introduction
2. Expected value of perfect information
3. Expected value of partial perfect information
1. Computation
2. Notes on EVPPI
4. Expected value of sample information
1. Computation
5. Expected net benefit of sampling
6. Summary key points
7. Further reading
8. Exercises
• References

• Appendix 1 Abbreviations
• Appendix 2 Common distributions
1. The Normal distribution
2. The Binomial distribution
3. The Multinomial distribution
4. The Uniform distribution
5. The Exponential distribution
6. The Gamma distribution
7. The Beta distribution
8. The Dirichlet distribution
• Index